refactor(extractors): added config in high-order extraction methods (#4)

CorentinTh · web-flow · commit 2a8b88e48a81 · 2025-06-29T14:08:11.000Z
diff --git a/src/extractors.usecases.ts b/src/extractors.usecases.ts
@@ -1,8 +1,8 @@
-import type { ExtractorConfig } from './types';
+import type { PartialExtractorConfig } from './types';
 import { parseConfig } from './config';
 import { getExtractor } from './extractors.registry';
 
-export async function extractText({ arrayBuffer, mimeType, config: rawConfig }: { arrayBuffer: ArrayBuffer; mimeType: string; config?: ExtractorConfig }): Promise<{
+export async function extractText({ arrayBuffer, mimeType, config: rawConfig }: { arrayBuffer: ArrayBuffer; mimeType: string; config?: PartialExtractorConfig }): Promise<{
   extractorName: string | undefined;
   textContent: string | undefined;
   error?: Error;
@@ -33,13 +33,13 @@ export async function extractText({ arrayBuffer, mimeType, config: rawConfig }:
   }
 }
 
-export async function extractTextFromBlob({ blob }: { blob: Blob }) {
+export async function extractTextFromBlob({ blob, config }: { blob: Blob; config?: PartialExtractorConfig }) {
   const arrayBuffer = await blob.arrayBuffer();
   const mimeType = blob.type;
 
-  return extractText({ arrayBuffer, mimeType });
+  return extractText({ arrayBuffer, mimeType, config });
 }
 
-export async function extractTextFromFile({ file }: { file: File }) {
-  return extractTextFromBlob({ blob: file });
+export async function extractTextFromFile({ file, config }: { file: File; config?: PartialExtractorConfig }) {
+  return extractTextFromBlob({ blob: file, config });
 }

Original file line number	Diff line number	Diff line change
`@@ -1,8 +1,8 @@`
`1`		`-import type { ExtractorConfig } from './types';`
	`1`	`+import type { PartialExtractorConfig } from './types';`
`2`	`2`	`import { parseConfig } from './config';`
`3`	`3`	`import { getExtractor } from './extractors.registry';`
`4`	`4`
`5`		`-export async function extractText({ arrayBuffer, mimeType, config: rawConfig }: { arrayBuffer: ArrayBuffer; mimeType: string; config?: ExtractorConfig }): Promise<{`
	`5`	`+export async function extractText({ arrayBuffer, mimeType, config: rawConfig }: { arrayBuffer: ArrayBuffer; mimeType: string; config?: PartialExtractorConfig }): Promise<{`
`6`	`6`	`extractorName: string \| undefined;`
`7`	`7`	`textContent: string \| undefined;`
`8`	`8`	`error?: Error;`
`@@ -33,13 +33,13 @@ export async function extractText({ arrayBuffer, mimeType, config: rawConfig }:`
`33`	`33`	`}`
`34`	`34`	`}`
`35`	`35`
`36`		`-export async function extractTextFromBlob({ blob }: { blob: Blob }) {`
	`36`	`+export async function extractTextFromBlob({ blob, config }: { blob: Blob; config?: PartialExtractorConfig }) {`
`37`	`37`	`const arrayBuffer = await blob.arrayBuffer();`
`38`	`38`	`const mimeType = blob.type;`
`39`	`39`
`40`		`- return extractText({ arrayBuffer, mimeType });`
	`40`	`+ return extractText({ arrayBuffer, mimeType, config });`
`41`	`41`	`}`
`42`	`42`
`43`		`-export async function extractTextFromFile({ file }: { file: File }) {`
`44`		`- return extractTextFromBlob({ blob: file });`
	`43`	`+export async function extractTextFromFile({ file, config }: { file: File; config?: PartialExtractorConfig }) {`
	`44`	`+ return extractTextFromBlob({ blob: file, config });`
`45`	`45`	`}`